1. Field of the Invention
The invention, in general, relates to a novel method of fabricating ion implanted doping layers in semiconductor layers and, more particularly, to a method of the kind referred to utilizing ultrasonic techniques to produce implantation layers of an extremely shallow depth of penetration.
2. The Prior Art.
Semiconductor materials such as silicon, gallium arsenide and gallium phosphide are used on a large scale for the fabrication of semiconductor devices. In a manner typical of modern semiconductor technologies, electrically active areas are fabricated in a substrate material during the course of their fabrication process, by imbedding p- and n-dopants by ion implantation.
Even today, the advance of modern semiconductor technology for the production of highly integrated switching circuits is to a substantial measure dependent upon the fabrication of extremely small electrically active areas and extremely flat or shallow junctions. The furthest developed demands are those in the field of silicon-based microelectronics. While the emphasis will hereafter be placed upon this field, it will nevertheless be understood by those skilled in the art that the discussion and its conclusions may, with certain modifications, be broadened to other semiconductor materials. The relevant demands on the silicon-based technology have heretofore been well described by the xe2x80x9cSIA Roadmapxe2x80x9d (National Technology Roadmap for Semiconductors; Semiconductor Industry Association (SIA) (1997)) of silicon microelectronics. In accordance therewith, depths of pn-junctions between 50 nm and 120 nm have been demanded in the year 1998 for advanced highly integrated switching circuits; by 2010 they are to be reduced to 10 nm to 30 nm in order to keep pace with the predictable lateral scaling.
Such a reduction does not only require a lowering of the implantation energies during imbedding of the dopants into the semiconductor material, but also a reduction of the thermal budget in annealing the implanted profiles. In addition, for such shallow junctions the desired depth of penetration depends heavily upon the so-called transient enhanced diffusion (TED) influenced by non-balanced weight point defects of the kind which may be generated during the implantation process.
Further processes which may generate non-balanced weight point defects are, among others, the oxidation of silicon which leads to the release primarily of interstitial silicon atoms, and thermal processes in the presence of SiO2 sediments in the crystal mass.
In accordance with the SIA roadmap, the problem of TED in implantation processes has in latter years acquired special significance. A great deal of literature dedicated to this problem has been published, and international workshops and conventions are being organized to this end (such as the International Works hop on Measurement, Characterization and Modeling of Ultra-Shallow Doping Profiles in Semiconductors; (USJ""95, ""97, ""99) NC, USA).
Various attempts to reduce TED have become known from publications.
Using the transient enhanced diffusion of boron as an example, it was possible to demonstrate that an additional implantation of fluorine ions (see, for instance, D. Fan, J. M. Parks, R. J. Jaccodine; Appl. Phys. Lettr., 59 (10), 1212 (1991)) reduces TED in oxidation processes. The authors ascribed this effect to the inclusion of fluorine in the formed SiO2 micro precipitates which prevent an injection of interstitial atoms during the growths of the precipitate. However, this variant may only be applied if the source for the Sii is constituted by the growing SiO2 precipitates.
Another attempt to minimize TED consists of forming a layer of silicon nitride on the silicon surface (see, for instance, S. Matsumoto, K. Osada, et al. in Defect and Diffusion Forum, v. 153-155, 25 (1998)). In this layer, tensile stresses are generated during tempering processes which, in turn, lead to pressure stresses in the Si layer close to the surface. This causes the formation of excess point defects xe2x80x94vacancies in the case at handxe2x80x94which can react in an enhanced manner with the Sii and thus lead to a reduction of Sii supersaturation. It is a substantial drawback of this process that the generated mechanical stresses may lead to a plastic deformation of the Si during the heat treatment. The resultant displacements are extremely detrimental as regards the properties of the transistors and switching circuits to be fabricated.
A certain reduction of TED may be achieved by an in situ photo stimulation by UV light during the boron implantation (see, for instance, J. Ravi, Yu. Erokin et al., Appl. Phys. Lett., 67 (15), 2158 (1995)). During implantation the sample was irradiated with a mercury lamp at an energy level of 35 keV and a dose of 5xc3x971014 cmxe2x88x922 at a temperature of 177xc2x0 K. During an ensuing tempering at 800xc2x0 C., it was found that the depth of penetration had been reduced by about 30 nm.
The disadvantage of this process is its relatively low effect (only about 30 nm) as well as the required effort for cooling the sample during the implantation.
A number of proposals for reducing TED relate to the insertion of carbon into the silicon lattice. The carbon reacts with the Sii and can reduce its super saturation. (see, for instance, R. Scholz, U. Goesele, J. Y. Huh, T. Y. Tan; Appl. Phys. Lett., 72, 200 (1998)). Tests involving delta doping in Si and SiGe demonstrated (see, for instance, P. A. Stolk, H. J. Gossmann, D. J. Eaglesham; J. Appl. Phys., 81, 6031 (1997) that TED may be reduced by a factor of 150 by an insertion of carbon at a concentration of 1xc3x971020 cmxe2x88x923. An additional implantation of carbon at a dose of 2.5xc3x971014 cmxe2x88x922 also reduces TED (see, for instance, N. E. B. Cowern, A. Caccieto, J. S. Custer, et al., Appl. Phys. Lett., 68, 1150, (1996)). It was found that the Sii was reduced by about 1.5 per implanted carbon atom. However, the disadvantage of this process is the necessary incorporation of an additional component (carbon) as well as a possible reaction of the carbon with the boron dopant which may lead to a partial electrical deactivation of the boron atoms. Further disadvantages relate to a deterioration of the transport properties of the Si and stimulated growth of SiO2 precipitates because of the carbon.
Nowadays, ultrasonic processes are used widely in technology, in particular for cleaning surfaces and for modifying structures of powders. Thus, S. Ramesh et al. (S. Ramesh, Y. Koltypin, A. Gedenken, J. Mat. Res., 12 (12) 3271 (1997) make use of ultra sound at an energy level of 100 W/cm2 and a frequency of 20 kHz during the production of amorphous SiO2 spheres. The cavitation effects lead to a change in the size and structure of the particles.
A method of treating the back side of Si wafers by means of ultrasonically vibrating metallic balls is disclosed by U.S. Pat. No. 4,018,626. The defects formed thereby in the back side of the silicon wafer act as a gettering sink for the point defects generated at the front surface of the wafer and improve the parameters of the semiconductor device fabricated on the wafer.
Using ultrasonic treatment at frequencies between 0.59 MHz and 1.2 MHz, it was possible in layers of polycrystalline SiO2 to stimulate a dissociation of Fei-B pairs and to extend the diffusion length of minority life expectancies (see S. Ostapenko, L. Jastrzebski, B. Sopori, Semicond. Sci. Technol., 10, 1494 (1995)).
None of these applications of ultrasound in semiconductor technology does, however, refer to solving the problem of fabricating ion implanted doping layers and to the generation of implantation layers of an extremely shallow penetration depth.
It is an object of the invention to provide a markedly improved method of fabricating ion implanted doping layers in semiconductor materials which while preventing transient enhanced diffusion makes it possible to produce extremely shallow doping profiles for use in the fabrication of semiconductor devices by large-scale industrial processes.
A further object of the invention resides in the provision of a method of the kind referred to of reducing, in a qualitatively different manner, TED during the implantation process by exertion of a deliberate effect, in situ, on the implantation process.
Further objects will in part be obvious and will in part appear hereinafter.
In the accomplishment of these and other objects, the invention, in preferred embodiments thereof, provides for reducing the transient enhanced diffusion typically occurring during the implantation process by subjecting the semiconductor crystal, during the implantation process, to the effect of ultrasonic oscillations at a power of between 50 W/m2 and 104 W/m2 in a frequency range between 0.01 MHz and 100 MHz.
During this process, the back side of the semiconductor crystal is affixed to a piezo element, such as, for instance lithium niobate, cadmium sulfide or piezo-ceramic material, by way of an acoustic link, for instance an oil suitable for hard vacuum treatment, and is then introduced into the implanting chamber. Directly during the implantation process the semiconductor crystal is subjected to ultrasonic vibrations at a power of between 102W/m2 and 104W/m2 in a frequency range from 0.01 MHz to 100 MHz. The introduction of ultrasound does not lead to any significant increase in the temperature of the Si wafer. Following implantation, tempering of the samples takes place in a temperature range necessary for the electrical activation of the dopant.
The depth distribution of the impurities in the semiconductor crystal with or without ultrasonic treatment can be measured by secondary ion mass spectrometry (SIMS). In addition, the samples were classified as to their homogeneity by measuring the electrical resistance of their layers.
The advantage derived from the invention resides in the fact that it is possible directly during the implantation process to suppress the formation of highly mobile doping atoms during a subsequent temperature treatment.
The method in accordance with the invention may be practiced on an industrial scale in modern implantation equipment and is thus suitable for economically solving the problem of TED for modern circuit generations.